Solution of the Load Flow Problem

Load flow solutions in power networks are solved to answer the following questions.

- What are the line transformer and generator loads
- What are the voltage levels throughout the system
- What are the effects of changes to the system
- Can a balanced steady state be reached

Power systems can not be modelled by the conventional circuit loop or nodal analysis
methods as loads are given in terms of complex powers and in steady state the
generators can be considered to be power sources. In general the load VARS (P
L
and
Q
L
) are known and some generator VARS (P
G
and Q
G
) are known and so a solution is
sought for the transmission and the remainder generator VARS (there must be at least
one unknown generator VAR in the system).

The problem is basically that of solving 2n non-linear algebraic equations for an n bus
system and therefore requires numerical analysis techniques. The standard techniques
used are:
- Gauss- Siedel
- Newton-Raphson

The Newton-Raphson technique is more popular as it is more computationally
efficient.

Figure 1 shows an example 4 bus system with four generators and loads linked by a
transmission network. The impedances shown are due to component impedances such
as transformers, transmission lines and generators.



Figure 1 Example 4 bus network

Z
1,2

= 1/Y
1,2

Z
1,3
= 1/Y
1,3

Z
3,4

= 1/Y
3,4

Z
2,4
= 1/Y
2,4


S
G1
, S
L1
, V
1
S
G2
, S
L2
, V
2
S
G3
, S
L3
, V
3
S
G4
, S
L4
, V
4
I
1
I
2
I
4
I
3
Basic Considerations

Transmission lines are passive and can be treated either as admittances or impedances
using n port network theory. The admittance approach has been found to be the more
suitable for the power flow problem. In the admittance approach the transmission
currents at the nodes are given by the node voltages and transmission system
admittances as:
| |
t n
= ( (

I Y V
(1)

where V
n
are the nodal voltage phasors ( =V
n

n
) and [Y] is the bus admittance
matrix (=Y ) the elements of [Y] are given by:

Diagonal elements of the matrix:
y
i,i
= all admittances connected to node i
(e.g y
11
= Y
1
+Y
1,2
+ Y
1,3
for Fig.1.)

Off diagonal elements of the matrix:
y
i,j
= - admittances connected directly between nodes i and j
(e.g y
1,2
= -Y
1,2
for Fig.1.)

Example

Figure 2 example system where all parameters are in per unit quantities

For the example system given in Figure 2. the following bus admittance matrix can be
constructed

| |
50 100 100 50 0
100 100 100 0 100
50 0 100 50 100
0 100 100 100 100
Y j
+ (
(
+
(
=
( +
(
+



Z
1,2
= j0.01
Z
1,3
= j0.02
Z
3,4
= j0.01
Z
2,4
= j0.01

1
1 1 = Z V
o

2
1 1 = Z V
o

3
1 0 = Z V
o


4
1 0 = Z V
o

I
1
I
2
I
4
I
3
The bus currents are then given by
| |
1
180
1
180
0.873 0.5
150 100 50 0 1
100 200 0 100 1.745 0.5
1
50 0 150 100
0.873 0.5
1
0 100 100 200
1.745 0.5
1
(
(
Z
( (
(
( (
Z (
( (
= =
(
( (
Z
(
( (
(

(
Z

(

I
j
o
o
j
o
o
e
j
e
t
t



Consider the currents at a node i as shown in Figure 3.

Figure 3. Fundamental node i or bus on a transmission system (subscripts
G=generation, T=transmission and L = load)


From Kirchhoffs current law we have
Gi Li Ti
= + I I I
(2)
or we can also write
* * *
Gi Li Ti
= + I I I
(3)
Multiplying by the bus voltage V
i
gives
* * *
Gi Li Ti
i i i
= + VI VI VI
(4)
Equation (4) is just the power balance equation which can be expressed as
Gi Li Ti
= + S S S
(5)
Splitting equation (5) into real and reactive parts gives

Gi Li Ti
P P P = +
(6)
Gi Li Ti
Q Q Q = +
(7)
}I
Ti
Transmission
I
Li
Load
I
Gi
Generation
Bus i

There are then 2 equations which have to be solved and equations (6) and (7) have
potentially 6 unknowns. Thus for a n bus system there could potentially be 6n
unknowns and 2n equations to be solved. In order to be able to solve the equations the
unknowns have to be reduced to 2 unknowns per bus or node.

The transmitted real and reactive power (P
Ti
and Q
Ti
) can be expressed in terms of the
bus voltage phasors (amplitude and phase) as from equation (1) we can put
,
1
n
Ti j
j =
=

i j
I y V
(8)

From equation (8) it then follows that:
,
* * *
1
Ti j
n
Ti i i
j =
= =

i j
S VI V y V
(9)
Or using y = ye
j
we can write

,
,
1
( )
=
= Z

S
i j j
n
Ti i i j i j
j
y VV o o
(10)

Equation 10 can be expressed in terms of real and reactive components as:
,
,
1
cos( )
i j
n
Ti i j i j i j
j
P y VV o o
=
=

(11)
,
,
1
sin( )
i j
n
Ti i j i j i j
j
Q y VV o o
=
=

(12)
Substituting equations (11) and (12) into equations (6) and (7) then reduces the power
flow problem to that of solving the following set of equations at each bus or node on
the system.
,
,
1
cos( )
i j j
n
Gi Li i i j i j
j
P P y VV o o
=
= +

(13)
,
,
1
sin( )
i j j
n
Gi Li i i j i j
j
Q Q y VV o o
=
= +

(14)

In order to solve the equations the number of unknowns have to be reduced to 2 per
bus or node. Normally all the load powers are known and the following table of bus
or node types defines which other unknowns are defined.

Bus Type Unknowns Knowns Approximate number
Slack P
G
and Q
G
V =1 p.u and = 0 At least 1
Load V and P
G
and Q
G
85%
Generator Q
G
and P
G
and V 15%

The slack bus is the reference bus which is normally set to have a voltage magnitude
of 1 p.u. and zero voltage phase angles. All other phase angles are with respect to the
slack bus phase angle. The slack bus has also undefined generation so that any
mismatch between supply and loss (losses can not be predicted before a solution is
known) can be absorbed in the slack bus for a solution to be found. The power losses
can be distributed between several generators and this is known as a distributed slack
bus. There are then 2n unknown and 2n equations to be solved in a n bus system and a
solution is then sought for all the unknown variables.